Old Second National Bank

Bank Liquidity Coverage
Ratio
Alex Lowe
Old Second National Bank
2
Rationale for Implementing LCR-Based Liquidity
Monitoring
• Old system was constructed before crisis with emphasis on Federal Reserve
as a source of funds
• Prior methodology primarily based on projections as opposed to stress tests
• Quantitative nature of LCR appeared as an advancement over loose
guidance prescribed by previous regulations
• Its not uncommon for regulations applied to larger banks to get applied to
smaller banks in some form or fashion
3
Should a Non-LCR Bank Implement the LCR as
Written?
• Asset classified as High Quality are extremely limited
• Achieving robust income in the securities portfolio becomes very
challenging
• Stressed deposit outflow assumptions are not specific to an individual
institution
• Loan funding assumptions also aren't specific to an individual institution
• For most non-LCR banks, implementing the Modified LCR as written would
not be ideal
4
Scenarios
• OCC requested that multiple stress scenarios be incorporated
▫ This request is consistent with applicable regulatory guidance
▫ They did not provide suggestions about what the scenarios should be
• LCR is meant to be a theoretical worst case scenario
▫ It was deemed base LCR should be harshest of all scenarios
• Scenarios Decided:
▫ Bank-Only Stress
▫ Economic-Only Stress
▫ Combined Stress (LCR equivalent)
• All policy limits based on combined BLCR scenario
5
Old Second's Bank Liquidity Coverage Ratio
(BLCR) Summary
• Designed off of Modified LCR, as opposed to LCR
• Largest deviation from regulation is what is defined as HQLA
• Minor changes to official deposit outflow assumptions
• Loan outflows were changed to be specific to institution
6
HQLA Deviations
High-Quality Liquid Assets
Modified LCR
O2 LCR
Treasuries
All Securities Judged to be Saleable
Federally Related Institution Securities
FHLB Borrowing Availability
Government Sponsored Enterprise Securities
(Limited)
50% of Over-Pledged Securities
Non-Financial Corporate Bonds (Very
Limited)
Municipal Securities (Extremely Limited)
•
Almost any marketable security should have some liquidation value even under the
most stressed circumstances
•
The OCC was supportive of including all asset classes as HQLA
•
BLCR includes the option to exclude any assets as HQLAs which are deemed
unsaleable
7
How to Establish Stressed Market Values for
Unencumbered Securities
• Should a haircut methodology be utilized?
• Benefits:
▫ Method is used in official LCR to value Level 2A and 2B assets
▫ Process is extremely simple, as it only entails simple multiplication
▫ Calculation results are transparent due to simplicity of calculation
8
Detriments of Using a Haircut Methodology:
Example
• Assume 10-year corporate spreads varied between 100 basis points and 400
basis points over a relatively long lookback period
• Also assume the current 10-year swap rate is 2.00% and you own a
hypothetical 10-year corporate bond with a 4.00% coupon
Theoretical Pricing Based on Historical Spreads
Price Extrema
Price
Best Price (100 bp spread)
$108.58
Worst Price (400 bp spread)
$85.12
Scenario Analysis Using Haircuts
Scenarios
Starting
Spread
Starting
10%
15%
20%
25%
Price
Haircut Haircut Haircut Haircut
Bad Recession
380 bps
$86.49
$77.84
$73.51
$69.19
$64.86
Economic Expansion
120 bps
$106.80
$96.12
$90.78
$85.44
$80.10
9
Discounted Cash Flow Analysis, the Superior
Choice
• A much better methodology is to establish stressed market spreads unique
to each security class
• A stressed valuation for each asset can then be established by utilizing a
discounted cash flow analysis
• Stressed Market Spreads can be fed into ALM 6, and the software will
determine a market value for each unencumbered security based upon
current market interest rates
• The results will never be starting-point dependent
10
How Stressed Discount Spreads were Determined
• Established by analyzing numerous security curves and indices acquired
from various sources
▫ Bloomberg
▫ Vendor and broker contacts
▫ Historical ALM discount spreads
• Based upon an OCC recommendation, all credit instruments were assumed
to experience a full letter rating downgrade
• Spreads before two months after the release of FAS 157 were disregarded
• FAS 157 - This enhancement allowed financial firms to report the market
value of their assets based on a hypothetical transaction in an orderly
market, as opposed to being forced to calculate market values based on a
forced liquidation or sold under duress scenario.
11
BLCR Stressed Securities Discount Spreads
FAS 157
April 2009
Historical Market Spreads
7000
6000
Stressed Securities Discount Spreads
12.97%
CLOs
5.17%
Corporates
3.44%
FFELP
6.57%
Private Label
1.00%
Agency MBS
Basis Points
5000
4000
3000
2000
1000
0
12/31/2008
12/31/2009
12/31/2010
BBB CLOs
A CLOs
12/31/2011
BBB Corp
12/31/2012
FFELP
12/31/2013
12
Factors in Calculating Stressed FHLB Availability
• Unencumbered securities collateral removed and sold
• Factor in reduction in loan balances due to credit migration (most difficult
aspect)
▫ 20% of residential loans pledged become ineligible (based on historical data
on loans pledged)
▫ 11% of commercial loans pledged become ineligible (estimated from
historical non-performing loan data)
▫ Haircut methodology is not ideal, but it is conservative
• Identical to securities, employ stressed market spreads to calculate loan market
values (ALM6)
• Bank Risk Rating of 5 is assumed, which causes the harshest FHLB haircuts
• Once all these factors are accounted for the current outstanding borrowings are
deducted from the total stressed availability calculated. Any remaining
availability is considered an HQLA.
13
Historical Loan Discount Spreads
700
600
Basis Points
500
Stressed Loan Discount Spreads
9.35%
Consumer Loan
4.55%
Home Equity Loan
18.00%
Overdraft
4.95%
C&I Loan
4.30%
C&I Tax Exempt Loan
6.00%
Construction Loan
5.45%
CRE Loan
4.80%
CRE Tax Exempt Loan
2.75%
Residential Real Estate Loan
400
300
200
100
0
12/31/2008
12/31/2009
Fixed RRE Bank Loans
•
12/31/2010
12/31/2011
Fixed CRE Bank Loans
12/31/2012
ARM RRE Bank Loans
Applying stressed spreads across the board is conservative, as it’s unlikely all
loans would have deteriorated enough to warrant such action.
12/31/2013
14
FHLB Haircuts
FHLB Base Haircuts
Risk Ratings
1
2
3
4
5
5%
5%
5%
5%
10%
20%
20%
20%
40%
50%
Agency MBS
2%
2%
2%
2%
5%
1-4 Family Loans
12%
12%
12%
25%
30%
ABS
Bank CRE Loans
FHLB Haircuts Assuming 25% Ineligibility Factor
Risk Ratings
1
2
3
4
5
ABS
28.75%
28.75%
28.75%
28.75%
32.50%
Bank CRE Loans
40.00%
40.00%
40.00%
55.00%
62.50%
Agency MBS
26.50%
26.50%
26.50%
26.50%
28.75%
1-4 Family Loans
34.00%
34.00%
34.00%
43.75%
47.50%
•
In our last BLCR measurement, total availability provided by pledged loans at
the FHLB was 53.19% less in the combined scenario than it was on the actual
calculation date.
15
Outflow Rates (the denominator)
Deposit Outflows:
• Outflow rates that are consistent with the official regulation:
▫ 3% for Stable Retail
▫ 10% for Other Retail
▫ 40% for Wholesale
• Outflow rates inconsistent with official regulation:
▫ LCR stated outflows for collateralized deposits should be dependent on
collateral type backing the deposits, which was originally implemented
▫ Based upon an independent auditor’s advice, the BLCR was changed to
employ a flat 5% outflow rate
• Consistent with the official Modified LCR Regulation, net inflows equals
outflows minus inflows (with total inflows capped at 75% of outflows),
which is then multiplied by 70%
16
Deposit Account Designation Methodology
• The LCR defines deposit categories largely based on deposit insurance, but
unfortunately doesn’t do a great job defining at what level these rules apply
• For the BLCR, at a tax ID Level we made the following assumptions:
▫ If aggregate balances are less than $250k, each deposit account is
considered fully insured
▫ If aggregate balances are greater than $250k, all deposit accounts are
not considered fully insured
• BDI was utilized to aggregate total account balances by Tax ID to achieve
the above logic
• This method is a simplification of FDIC deposit insurance rules, but results
in a conservative assumption as it understates true balances insured
17
Loan Funding Outflows
• Regulation outflow rates were disregarded
• Utilizing Bank specific data appeared like a much more robust methodology
than applying constant prescribed outflow rates
• Administering constant rates also restricts flexibility in changing funding
rates when appropriate
• Commercial funding outflow percentages are either based on recent funding
history or lending estimates of funding probability
• Line of credit outflows are determined by applying largest historical
absolute percentage increase to each respective line type's availability
18
Inflows
• 50% of performing contractual payments from retail and small business
customers
• 50% of performing contractual payments from public funds and large
business customers
• 100% of contractual payments due to the bank on securities it owns that are
not eligible HQLA
• For all scenarios, inflows are capped at 75% of outflows, assuring there will
always be net cash outflows
• Inflows generally act as only a minor offset to the very harsh outflow
assumptions applied
19
Policy Limits
• Since the system is still becoming vetted, policy limits were set
conservatively:
▫ Guidance Limit is 1.50
▫ Liquidity Action Plan activated at 1.25
▫ Absolute Limit is 1.00
• As the ratio moves and refinements are potentially made, changes to the
limits will be made if deemed appropriate
20
The BLCR Ratio
• Results over the last few months have varied between 2.91 and 3.42
• Moderate volatility month to month primarily due to:
▫ Ebb and flow from loan funding and deposit movements
▫ Changes in composition of account types (stable retail, other retail, etc.)
▫ Expected P&I inflow differences between months
21
Sources/Uses Six Month Projection
• At OCC request a 6-month forward sources/uses projection was also
implemented
• The net inflow or outflow is added to the denominator to establish a new
BLCR figure for 6 months
• Methodology generally produces a timing mismatch for cash flows
• Emphasis is not being placed on the results of these projections
• Process does, however, put any large cash flow items on the radar screen
22
Other Scenarios
Six total base BLCR calculations:
• FHLB Available
▫ Bank-Specific
▫ Economic-Specific
▫ Combined
• FHLB Unavailable
▫ Bank-Specific
▫ Economic-Specific
▫ Combined
23
HQLAs for All Scenarios
Scenario Discount Spreads
Stress Scenario
•
•
*
Asset Type
Economic Downturn
Bank Specific
Combined
Securities
Stressed
Normalized
Stressed
Bank Loans
Normalized
Stressed
Stressed
Bank loan values not impacted in economic-only stress
Securities values not impacted in bank-only stress
Normalized
Stressed
24
FHLB Availability for All Scenarios
Scenario FHLB Haircuts
Stress Scenario
•
•
•
*
Asset Type
Economic Downturn
Bank Specific
Combined
Bank CRE Loans
20%
50%
50%
1-4 Family Loans
12%
30%
30%
Bank FHLB Risk Rating of 2 assumed as normalized level
Bank FHLB Risk Rating of 5 assumed as stressed level
Bank Loan Values are also assumed stressed in bank-specific and combined scenarios
Normalized
Stressed
25
Outflows for All Scenarios
BLCR Outflow Percentage Summary
Outflow:
Economic
Bank Specific
Combined
Notes
1. Stable Retail and Small Business
1%
2%
3%
2. Other Retail and Large Business
3%
7%
10%
3.Wholesale
10%
30%
40%
4. HQLA Secured Wholesale and
Two-Way CDARS
2%
3%
5%
5. Contractually Under-Pledged
Amounts
Same
Same
Same
Relevant for Public Fund and
Repo Customers
6. Lines of Credits
Same
Same
Same
Based on Historical data
7. Commercial Loan Commitments
Same
Same
Same
Based on Historical data
8. Outstanding FHLB Borrowings
0%
0%
0%
Notes: Any entries that say "Same" mean the outflow percentage is calculated and is the same
for all scenarios.
26
Other BLCR Considerations
• Determining deposit outflow percentages based on institution-specific
historical data
▫ Methodology ultimately wasn’t incorporated into model due to strategic
balance sheet runoff during the crisis
▫ Differentiating between organic deposit outflows and strategic outflows
proved too challenging
• Incorporating interest rate movements into scenarios
▫ Incorporating rate movements would have resulted in a less conservative
assumption than keeping rates static
• Creating entire rate specific scenario(s)
▫ Challenging to determine what should assumptions should correspond
to rate movement scenarios
27
Considerations for Other Banks
• Larger banks might consider stressing securities market spreads more due
to potential effects of trading illiquid bonds in large volume
• Incorporate other HQLAs deemed appropriate:
▫ Loan Sales
▫ Securitizations
▫ Brokered Deposits
• Include additional HQLAs stemming into the future:
▫ Additional FHLB/FRB availability projected from loans expected
available for pledging in the future
▫ Market value of securities pledged to public funds customers that would
accept an FHLB Letter of Credit
28
Bank LCR Example
FHLB Available Assumption
1-Month Timeframe
Baseline
Scenario
Numerator Components:
Overnight Investments
Unpledged Securities
Releaseable Public Fund Securities
Net FHLB Availability From Loans
Releaseable Securities from FHLB
Total HQLA
Denominator Components:
Scheduled Loan P&I Inflow
Scheduled Securities Inflow
Stable Retail Outflow
Other Retail Outflow
Wholesale Outflow
Public Funds and Repo Outflow
FHLB Outflow
LOC Increases Outflow
Loan Commitment Outflow
Other Outflows
Net Outflow Amount*
Liquidity Coverage Ratio
Guidance Limit
Absolute Limit
46,949,904
2,626,496,149
256,228,542
644,856,733
1,062,283,114
4,636,814,443
Rising Rate
Scenario
Economic
Stress
Scenario
Bank-Specific
Combined
Scenario
Stress Scenario
46,949,904
46,949,904
46,949,904
2,466,542,534 2,063,893,682 2,617,735,832
240,624,224
256,228,542
256,228,542
605,584,958
551,752,794
152,893,510
997,590,073
938,347,837 1,076,125,500
4,357,291,692 3,857,172,760 4,149,933,288
46,949,904
2,063,893,682
256,228,542
152,893,510
938,347,837
3,458,313,476
(116,361,158) (116,361,158) (116,361,158)
(13,918,267) (13,918,267)
(13,918,267)
85,239,317
170,478,634
255,717,951
117,061,773
273,144,136
390,205,908
105,693,296
317,079,887
422,773,183
63,242,461
63,242,461
63,242,461
217,017,170
24,113,019
241,130,189
255,364,781
255,364,781
255,364,781
499,337,561
681,200,445 1,048,708,534
7.72
6.09
3.30
1.50
1.00
29
Bank LCR Example Cont.
FHLB Unavailable Assumption
1-Month Timeframe
Baseline
Scenario
Numerator Components:
Overnight Investments
Unpledged Securities
Releaseable Public Fund Securities
Releaseable FHLB Securities
Total HQLA
Denominator Components:
Scheduled Loan P&I Inflow
Scheduled Securities Inflow
Stable Retail Outflow
Other Retail Outflow
Wholesale Outflow
Public Funds and Repo Outflow
FHLB Outflow
LOC Increases Outflow
Loan Commitment Outflow
Other Outflows
Net Outflow Amount*
Liquidity Coverage Ratio
46,949,904
2,626,496,149
256,228,542
1,062,283,114
3,991,957,710
Rising Rate
Scenario
Economic
Stress
Scenario
Bank-Specific
Combined
Scenario
Stress Scenario
46,949,904
46,949,904
46,949,904
2,466,542,534 2,063,893,682 2,617,735,832
240,624,224
256,228,542
256,228,542
997,590,073
938,347,837 1,076,125,500
3,751,706,735 3,305,419,966 3,997,039,778
46,949,904
2,063,893,682
256,228,542
938,347,837
3,305,419,966
(116,361,158) (116,361,158) (116,361,158)
(13,918,267) (13,918,267)
(13,918,267)
85,239,317
170,478,634
255,717,951
117,061,773
273,144,136
390,205,908
105,693,296
317,079,887
422,773,183
63,242,461
63,242,461
63,242,461
217,017,170
24,113,019
241,130,189
255,364,781
255,364,781
255,364,781
499,337,561
681,200,445 1,048,708,534
6.62
5.87
*Consistent with the official Modified LCR Regulation, net inflows equals outflows minus inflows (with total inflows capped at 75% of
outflows), which is then multiplied by 70%
3.15
30
Back-Testing
• Old Second management requested that the new LCR system be back-tested
to evaluate the system’s effectiveness
• October 2010 was chosen because it appeared this was when the bank was
in its weakest historical liquidity position
• With some slight assumption variations due to data restrictions, the
ultimate result calculated was 0.85.
• Had the BLCR been in place at the time, the LAP plan would have been
activated well before this point, thus buffering liquidity and reducing risk
• New system provides more robust liquidity monitoring and is more
conservative than old methodology
31
Conclusion
• The actual LCR is a long and nuanced regulation
• With appropriate modifications it can be made vastly less complex
• Reasons for implementing an LCR system include:
▫ It incorporates stress in all major aspects of the balance sheet: Inflows,
Outflows, and Asset Valuation
▫ It is generally forward looking stress test, as scenario would not be
expected to occur overnight
▫ Regulatory Kudos